Performance Comparison of Instrumentation Amplifiers

61 downloads 0 Views 589KB Size Report
Instrumentation Amplifier – Difference Amplifier, 3 Stage IA, IA with linear gain - for ... Ratio of Differential mode gain to common mode gain is called Common ...
ICDATPM-2014

ISBN:978-81-924212-6-1

Performance Comparison of Instrumentation Amplifiers – A Beginner’s View Bhargava Peddiraju1, Ravi Kumar Jatoth2, Nagaraju Duggirala3 1,3

Department of Electronics and Communication Engineering, NIT W, Warangal, India. 1

2

[email protected] 3 [email protected]

Department of Electronics and Communication Engineering, NIT W, Warangal, India. [email protected]

Abstract—Instrumentation amplifiers (IAs) have become inevitable in all kinds of sensitive instrumentation and control applications. For an Engineer, a brief and qualitative idea of how an instrumentation amplifier works will be useful in a multitude of ways. In this paper, we describe and suggest three different configurations and modifications that can be used to design an Instrumentation Amplifier – Difference Amplifier, 3 Stage IA, IA with linear gain - for various applications. Main theme of the paper is to describe the flaws and advantages of using these configurations with proper justification. Effort has been made to bring out various obvious implementation issues which greatly affect the final design as per applications. Keywords — Instrumentation; Amplifier; linear gain; CMRR; Temperature dependence.

I. INTRODUCTION Signals produced by transducers are very small levels of currents or voltages, often indistinguishable from noise. Instrumentation Amplifiers (IAs) are the back bones in Process control and Instrumentation applications, which help in sensing the quantity to be measured and bring them to a manageable level for further processing and controlling. Their usage as transducer signal conditioning circuits puts them often in harsh and rapidly changing environments with varying noise and interference from unwanted sources. So an IA must have the ability to amplify the desired signal and reject unwanted signals, intelligently in a sense. In this respect, an IA must have the following characteristics [1]: • Ability to separate signals from ambient noise. • Ability to amplify the signal as per a desired linear transfer characteristic. • Ability to sense signals from sensor and amplify it without loading it. • Insensitivity to ambient environmental changes and interferences. With these considerations in mind, we proceed to describe the three basic configurations – Difference Amplifier (DA), 3 stage Instrumentation Amplifier and Linear gain IA. Main aim

of the paper is to highlight the following aspects of Instrumentation Amplifier design: • A beginner has some trouble designing an IA. It is the case that most of the times, bread board implementations do not lead to the desired results unless some definite procedures are followed. Effort will be made to make fundamental concepts clear. • Effect of temperature on the overall performance of the IA will be elaborated with support from simulation results. • Resistance mismatches in the circuits commonly used cause unforeseen variations to the output. Such effects have been simulated and definite conclusions will be drawn from the results. We present different simulation results making some conclusions obvious regarding choosing a configuration. All these simulations have been done using NI Multisim® software and graphs plotted using MATLAB®. II. PRELIMINARIES Some introduction to the terminology associated with signal conditioning and very much related to an IA design is necessary to be highlighted, so as to make the rest of the paper easy to understand. These terms are the ones used by manufacturers as well, in their data sheets [1]. • Ratio of Output voltage to the input signal voltage when the input applied is not common to the input terminals of an IA is called Differential Gain (Ad). If the applied input is common to both the terminals, the ratio is called Common mode Gain (Acm). • Ratio of Differential mode gain to common mode gain is called Common Mode Rejection Ratio (CMRR). It is expressed in decibels as, CMRR (dB) = 20 log10 (Ad/Acm).

(1)

Other terms used in literature will be defined when they are encountered later in the paper. The three terms defined above provide very efficient metrics to qualify an IA.

124

ICDATPM-2014

ISBN:978-81-924212-6-1

III. THE DIFFERENCE AMPLIFIER The basic building block which gives an IA its ability to reject common mode noise is the Difference Amplifier. For applications which do not require high accuracy, a DA alone can do the job with a very simple circuit configuration, using a single operational amplifier (Op – Amp). The circuit is shown in Fig. 1. In a DA, we give one signal to the inverting terminal, to get an output given by Vout(i) = - (R2/R1) Vin,

(2)

and the other signal to the non-inverting terminal, because of which the output becomes Vout (ni)= [R4/(R3 + R4)] (1 + R2/R1) Vin.

(3)

are some of the factors that influence the output of the DA, as observed through simulations [5]. • The circuit provides appreciable common mode gain, which is because of different input impedances offered by the inverting and non-inverting terminals. • Also, the circuit is unstable when temperature and tolerance changes are encountered during circuit operation, as seen from the AC analysis. • It is observed that as long as the bridge is balanced, gain at room temperature remains fairly constant. • At very low voltages, peak in gain is observed, because of the offset voltage, still being amplified by the circuit. Since the circuit is solely producing gain, it makes offset errors much difficult to deal with. But, it is easy to design, less space consuming and low power consuming because of just one Op - Amp. The disadvantages being that it provides low CMRR and less stability with temperature and at high frequencies. Gain variation is very difficult to achieve because any variation in resistances will disturb the bridge balance and deteriorates the output. The solution comes from a simple modification of the circuit which becomes quite obvious as we go on. The modified circuit tends to be much better than the basic difference amplifier, making it the most popular configuration used in industry. The circuit is 3-Op Amp Instrumentation Amplifier, which has become synonymous with the name Instrumentation Amplifier. IV. 3 OP – AMP INSTRUMENTATION AMPLIFIER

Fig. 1 The Difference Amplifier (Gain = 100)

When we use the principle of superposition, the final output due to both the sources becomes Vout = Vout(I) + Vout(NI).

(4)

The circuit schematic diagram of the 3 – Op Amp Instrumentation Amplifier is shown in Fig. 2. A similar analysis done in the case of DA will help in getting the equations for designing this IA [2]. For the difference amplifier stage, we have again the condition,

Now, when the resistances are randomly selected, two signals are not amplified by equal amounts. This is catastrophic if the DA has to operate in the presence of Common mode noise. Hence, to avoid this problem, we impose the following condition [1][2], (R2/R1) = (R4/R3),

R2 / R1 = R4 / R3

(5)

which reduces the output equation to simply Vout = (R2/R1) (VNI – VI)

(6)

Where VI is the input given to inverting terminal and VNI is that given to non-inverting terminal, both referenced to a single reference voltage, for example, ground voltage. This specification of reference is important because the difference between the signals is going to be amplified here and this depends on the reference potential considered. A. Some Inherent Problematic Issues Though the DA seems deceptively simple, even the slightest of the changes can ruin its desired functionality. The following

Fig. 2 3 Op –Amp Instrumentation Amplifier

125

(7)

ICDATPM-2014

ISBN:978-81-924212-6-1

And when R5 = R6, the equation for output voltage becomes [2] V out = [1+ (2*R5 / Rg)] (R2/R1) (V in2 – V in1)

(8)

Now, going into the functioning, the circuit has two stages: The input Buffering Stage and the Difference Amplifier Stage. Input Buffering Stage: It helps in [1] • Providing high input impedance for both inverting and non-inverting terminals, to avoid loading. • Providing equal input impedances for both terminals to provide symmetry. • Increasing CMRR by doing buffering. • By avoiding offset errors of the Op Amps used in the input buffering stage by using dual Op Amp ICs in which Op Amps track each other closely and provides the gain stage as well. Differential input signals are amplified by the Equation (8) while common mode signals are rejected here itself. Common Mode signals are just buffered with unity gain by each of the two buffering Op Amps. At the resistance Rg, these voltages are subtracted. Difference Amplifier’s role is clear by now. It just makes the subtraction and removes the noise. Gain is not provided by this stage. Offset error of input buffer stage are compensated by using closely tracking Op Amps. Offset errors of difference amplifier stage can be minimized using the right combination of resistors. This stage does not add any gain, so errors can be set to a minimum always, since we are not touching this section even in case of gain variation requirements. Disadvantages include issues namely, the number of Op – Amps required, leading to more power consumption and area. Gain variation is possible by Rg, but it is not linear. To make the gain linear, another variation the design is used widely, as discussed in the next section.

The offset error of the additional op amp is responsible for some reduction in CMRR. When compared with previous design, this design provides linear gain variation, but also reduces the noise rejection capability. Linear gain variation with simple adjustment of Rg, is the only additional advantage this circuit provides over the generic 3 - stage Op Amp IA configuration. Using the analysis shown in [2], we have, under bridge balance condition, and R5 = R6, Vout = [1+(2*R5/Rx)][(R2/R1)*(Rg/Rf)](Vin2 – Vin1) (9) Apart providing the advantage of making gain linear with Rg, this circuit is disadvantageous in the sense that it needs additional Op – Amp for proper functionality, increasing the area and power requirements. The following sections show the simulation results and conclusions are drawn on the basis of the results in the graphs. VI. SIMULATIONS Before going to simulations, it must be noted that the Op – Amps used are the generic 741 models available in Multisim®. The model parameters are as listed here in Table I. It has to be noted that offset compensation is not done in these cases, to see the effect of such unwanted voltages on the outputs. TABLE I MODEL PARAMETERS FOR 741 OP – AMP IN SIMULATIONS S. No.

V. LINEAR GAIN INSTRUMENTATION AMPLIFIER The circuit schematic for this design is shown in Fig. 3. It is seen that addition of one more op amp in the feedback path of Difference Amplifier stage causes some effect in the operation.

Fig. 3 Instrumentation Amplifier with Linear Gain

Parameter

Value

1

Input Resistance

2 MΩ

2

Open Loop Gain

200000

3

Output Resistance

75 Ω

4

CMRR

90 dB

5

Input Offset Voltage (V os)

1 mV

6

Input Offset Current (I os)

20 nA

7

Input Bias Current (I bs)

80 nA

A. Difference Amplifier (DA) From Fig. 4, it is clear that component mismatches and variations over large ranges in DA can cause dramatic changes in gain non – linearly. Both common and differential mode gains follow a similar pattern but, common mode output voltage for the designed circuit is 11.2 while that of differential mode is 100. This makes CMRR equal to 19.015 dB. It is obvious that the performance is poor. To improve in this aspect, the following steps are suggested: • Maintaining R1 = R3; R2 = R4 will remove the offset voltage caused by Input Bias currents. • Internally compensated Op – Amps are to be used to remove the effect of Offset Voltages. If not possible, external offset compensation is to be adopted. • Components used must be very precise and sufficiently of high values.

126

ICDATPM-2014

ISBN:978-81-924212-6-1

Fig. 5 DA – Variation of Common and Differential mode gains with changes Temperature.

Fig. 4 DA – Variation of Common and Differential mode gains with changes in resistor R2.

• Because, at high resistance values, slight variations and mismatches in their values do not affect the output much. If low value resistors are used, slight changes in their resistances, even changes in the lengths of connecting wires will give rise to some common mode output which may not be acceptable to the operation. Fig. 5, shows the variation of gain with temperature for a DA. All resistor values shown in the circuit of Fig. 1, are at a temperature of 27 0C. It is clear that the differential gain does not change much and hence is stable for the given resistance temperature coefficients considered. But, the common mode gain changes more. Even a change of temperature by one degree can cause common mode gain to change by 3.5 %. To solve this problem to some extent, the following steps are suggested. • Resistors of very low temperature coefficients of resistance which are less vulnerable to temperature fluctuations are to be used. • Precision Op – Amps with temperature insensitive characteristics are to be used. • If both such cases are not possible then resistors can be replaced with potentiometers and frequent calibration is needed to remove the common mode noise. This is not practical in many situations. As far as AC analysis is concerned, the circuit follows a predictable variation and not much variation in terms of band width. The next section is about the 3 Op – Amp IA.





B. 3 Op – Amp Instrumentation Amplifier Fig. 6, shows the variation of gain of the IA shown in Fig. 2. It must be noted that the difference amplifier stage of this circuit is still subject to all the effects discussed in the previous sections. To minimize those effects in this case, we choose the gain of the difference amplifier stage to be unity. Some straight forward conclusions can be drawn from the comparison of gain curves for both DA and 3 Op – Amp IA. They are summarized below.

It is evident that the gain can be varied in a predictable but non – linear fashion with variation in Rg. With such variations, the common mode gain does not change. Even the offsets of the buffer Op – Amps appear as common mode voltage to the difference amplifier stage and they are rejected. Further external compensation and using dual Op – Amps for the input stages are good design methodologies to bring down the common mode gain further. The CMRR in this case if proportional to differential gain, which is desired characteristic. Changing Rg can take CMRR to as high as 90 or 100 dBs. Rg here is the gain setting element. A micro controller or an analog linearization circuit can be used to make the gain variation linear, which is desirable.

Fig. 6 IA – Variation of Common and Differential mode gains with changes in resistor Rg



127

These design considerations are to be coupled with those considered in DA implementation to get a very efficient IA.

ICDATPM-2014

ISBN:978-81-924212-6-1

Now, going to the temperature dependencies, we have the curve of Fig. 7. The following conclusions can be made out. • Common mode gain changes here by 0.7 % for one degree change in temperature for the same difference amplifier stage design. So, temperature stability is obviously better in this case. • It is observed that common and differential mode gains are almost equally changing and hence, effect on CMRR is very less with temperature. In fact, at the given normal temperature of 27 0C, CMRR of the circuit is 33.13 dB, for the given value of Rg equal to 1 KΩ. Further reduction of Rg will cause better CMRR.

Fig. 10 IA with linear gain – Variation of Common and Differential mode gains with Rg

Fig. 11, shows variation of gain with temperature for the circuit shown in Fig. 3. In this case, the common mode gain changes by 2.75 % per one degree change in temperature. Some disadvantages of DA appear here because of the inclusion of an uncompensated normal Op – Amp in the feedback circuit.

Fig. 7 IA - Variation of Common and Differential mode gains with changes Temperature

All these conclusions point to the fact that this configuration is far better and easily operable compared to the DA alone. Even in the case of AC operation, the phase and gain plots show a non-linear variation with increasing Rg and a linear variation with increase in temperature. With temperature increase, the bandwidth gets slightly reduced, similar to the DA case. C. 3 Op – Amp Instrumentation Amplifier with linear gain Fig. 8, shows the effect of changing Rg in the feedback element of the circuit shown in Fig. 3. This is the gain variation element. From the graph, it is observed that • Both common and differential mode gains are linear varying in nature with changing Rg. • But, CMRR here also changes because common mode noise is not constant but it also varies. The reason for this is that the Op – Amp sitting in the feedback part of the circuit injects some voltage of its own and hence spoils the CMRR. To avoid this, the feedback Op –Amp must be an ultra-low input current, compensated Op – Amp. • All the design methodologies followed in the previous sections must be followed here as well. But, it has to be observed that the feedback given here at the difference amplifier stage if positive. Because, the Op – Amp in the feedback network will invert the feedback signal automatically. In effect, we eventually end up obtaining a negative feedback.

Fig. 11 IA with linear gain – Variation of common mode and differential mode gains with temperature

It has to be noted that the same linear gain variant can be implemented with just a DA as well. The connection of feedback Op – Amp remains the same in both cases. In DA implementation, using this will improve the performance slightly. In this circuit implementation, there is a slight variation in AC response as well. At low values of gain, an additional pole is obtained in the response. At higher gains, this poles effect reduces and the response tends to be that which can be observed in the previous two cases. This additional effect is because of the inclusion of the Op – Amp in the feedback path. Care should be taken when high frequency applications demand the usage of this kind of circuit configuration.

128

ICDATPM-2014

ISBN:978-81-924212-6-1

VII. COMPARISONS AND CONCLUSIONS Table II, gives the comparative results of all the three configurations considered in this paper for Instrumentation Amplification applications. It is clear from the above table that the 3 Op – Amp IA has the best stability with temperature. DA has the least. Linear gain variant has an intermediate nature of stability. TABLE II COMPARISON OF GAIN VARIATIONS [For the circuits considered]

% Change with temperature (/0C) Common Differential Mode gain Mode gain

DA

3.5

1.11

0.7

0.7

2.75

0.67

3 Op – Amp IA IA with Linear Gain

Change with change in gain element (/KΩ) Common Differential Mode gain Mode gain Non Non Linear Linear Non 0 Linear 3.1677

140

The following application oriented conclusions can be made finally from the analysis done in this paper. • If space and power constraints are more important, with not-so-strict accuracy and precision limitations, then DA will be the best choice. But, care must still be taken to avoid drastic variations in performance, if the environment of working is dynamically varying. • If linearity is a concern and accuracy and noise rejection can be compromised a little, the linear gain variant can be used. Power and space may be compromised in this case. • If accuracy and noise rejection, coupled with robustness, is the only essential criterion, then the 3 Op - Amp IA can be used. Using a quad Op – Amp IC directly will save space. A trade-off has to be made with power consumption.

Other variations like the 2 Op – Amp IAs and linear circuits with T bridges are also possible to be implemented. The final decision is left to the designer on the basis of the application in hand and balancing the requirements while compensating for the losses and non-idealities. REFERENCES [1] Charles Kitchin and Lew Counts, A designer’s guide to Instrumentation Amplifiers, 3rd ed., Analog Devices, 2006, pp. 1.3 – 2.5. [2] Sergio Franco, Design with Operational Amplifiers and analog Integrated circuits, 3rd ed., Tata McGraw Hill, 2002, pp. 71 – 98. [3] J. H. Huijsing, “Instrumentation Amplifiers: A Comparative Study on Behalf of Monolithic Integration”, IEEE Transactions on Instrumentation and Measurement, Vol IM-25, No. 3, pp. 227-231, September 1976. [4] A. B. Grebene, Bipolar and MOS Analog Integrated Circuit Design, New York: John Wiley and Sons, 1984. [5] Analog Devices Engineering Staff, Practical Design techniques for Sensor Signal Conditioning, Analog Devices, 1999. [6] J. Wong and A. Gracia, Precision Transducer Interfaces Amplifier Applications Guide, Analog Devices, 1992. [7] J. R. Riskin, A User’s Guide to IC Instrumentation Amplifiers, Analog Devices, 1993. [8] Tobey, G. E., J. G. Graeme and Huelsman, Operational Amplifiers – Design and Applications, McGraw-Hill Book Company, 1971. [9] Schick, Larry L, “Linear Circuit Applications of Operational Amplifiers”, IEEE Spectrum, April 1970. [10] Barna A, Operational Amplifiers, John Wiley & Sons, Inc., 1971. [11] David F.Stout, Handbook of Operational Amplifier Circuit Design, McGraw-Hill Book Company, 1998, pp. 9-1 to 9-8.

129